Ranking 01
Cheapest
Lowest true cost per GPU-hour, in stock or stock unknown, newest reading per provider
OpenRelay1× · 21:11 UTC$2.25
Cyfuture AI8× · 21:10 UTC$3.04- 4
DigitalOcean1× · 21:10 UTC$3.80
AMD · CDNA 3 · Datacenter · 6 providers list itAMD Instinct MI325X cloud prices
Rented today by 4 providers with a current on-demand offer, from $2.25 to $3.80 per GPU-hour, median $2.65. Every number below links to the page it was read from, with its read date.
Memory256GB HBM3
Bandwidth6.00TB/s
FP16 tensor1,307TFLOPS
FP32163.4TFLOPS
Power1000W
Specifications: manufacturer figures.
Cheapest on-demand · read 21:35 UTC
$2.25/ GPU·h
TensorWaveTrust 648× MI325X · 256 GB- Median
- $2.65 4 providers, once each
- 30 days
- +1.5%
- 90 days
- +1.0%
Prices by rental mode · On-demand · Spot · ReservedMI325X prices today
Cheapest, median and spread among current offers. Medians count one floor per marketplace and need at least three offers.
Cheapest$2.25TensorWave
Median$2.65n = 7
Highest$3.80among current offers
Current offers94 providers · last 45 days
Ever listed175 providers on file
Where today's on-demand offers sitone dot per eligible offer (7) · the pool behind the tiles · read Nov 1, 2026 21:35 UTC
Provider offer7Median$2.65Middle half$2.65 – $3.12
7 eligible offers from 4 providers, binned every $0.05: from $2.25 (OpenRelay) to $3.80 (DigitalOcean). Hover a dot for the provider, price, configuration, region and read date; click it to open the provider.
On-demand price history · 33 daysdaily median across all providers · 1 provider lines from their 30-day series · read Nov 1, 2026
Median · all providers$2.65Providers
Today $2.6530 days +1.5%33 days +1.0%Low $2.65 · Oct 31High $3.11 · Oct 21
Daily median on-demand price per GPU-hour across all providers, 33 days to Nov 1, 2026. Provider lines: the provider's cheapest current on-demand offer, its listed price over the last 30 days (read Nov 1, 2026). Drag to zoom, double-click to reset.
Today's podium · on-demand · readings of the last 45 days · reference Nov 1, 2026 21:35 UTCWho rents the MI325X best today
Cheapest true cost, best throughput per dollar, highest Trust Score. One provider per step, ten places, every row dated.
Ranking 02
Best value
Most throughput per dollar for LLM inference · Trust ≥ 60 first, then provisional · estimated
- DigitalOcean1× · 21:10 UTCValue 73$3.80
- OpenRelay1× · 21:11 UTCValue 86prov.$2.25
- 4Cyfuture AI8× · 21:10 UTCValue 81prov.$3.04
Ranking 03
Most trusted
Highest Trust Score among providers with an eligible offer · unscored providers follow, cheapest first
- TensorWave8× · 21:11 UTCTrust 64$2.25
- OpenRelay1× · 21:11 UTCnot scored$2.25
- 4Cyfuture AI8× · 21:10 UTCnot scored$3.04
7 eligible on-demand offers from 4 providers · median $2.65 (n = 7) · not ranked: 2 unavailable · readings of the last 45 days · reference Nov 1, 2026 · 21:35 UTCRankings are never sponsored. How we rank →
All offers · 55 readings on file · 6 providersEvery MI325X offer, with its receipt
Current offers ranked by true cost, each with its read date. Older readings stay listed below, in grey, and never enter a ranking.
| Provider | Configuration | Region | Stock | List | True cost | Read | |
|---|---|---|---|---|---|---|---|
| OpenRelay | 1× MI325X | — | Stock unknown | $2.25 | $2.25 | 21:11 UTCfresh | Receipt |
| TensorWaveTrust 64 | 8× MI325X3,000 GB RAM · 15,360 GB disk | — | Stock unknown | $2.25 | $2.25 | 21:11 UTCfresh | Receipt |
| Cyfuture AI | 8× MI325X128 vCPU · 1,536 GB RAM | — | Stock unknown | $3.04 | $3.04 | 21:10 UTCfresh | Receipt |
| Cyfuture AI | 4× MI325X64 vCPU · 768 GB RAM | — | Stock unknown | $3.07 | $3.07 | 21:10 UTCfresh | Receipt |
| Cyfuture AI | 2× MI325X32 vCPU · 512 GB RAM | — | Stock unknown | $3.11 | $3.11 | 21:10 UTCfresh | Receipt |
| Cyfuture AI | 1× MI325X16 vCPU · 256 GB RAM | — | Stock unknown | $3.14 | $3.14 | 21:10 UTCfresh | Receipt |
| DigitalOceanTrust 82 | 1× MI325X | — | Stock unknown | $3.80 | $3.80 | 21:10 UTCfresh | Receipt |
 VultrTrust 75 | 8× MI325X252 vCPU · 2,872 GB RAM · 14,336 GB disk | no location can deploy it right now | Unavailable | $4.62 | $4.62 | 21:08 UTCfresh | Receipt |
| VultrTrust 75 | 8× MI325X256 vCPU · 3,072 GB RAM · 3,576 GB disk | no location can deploy it right now | Unavailable | $4.62 | $4.62 | 21:08 UTCfresh | Receipt |
9 fresh offers · 8 older readings of the same provider replaced by a newer one · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
Value · what the hour buysPrice per unit, and what fits
The same hour, divided by what the card gives you; then the smallest real configuration for the most downloaded open models.
| Unit | At the cheapest | At the median |
|---|---|---|
| per GB of VRAM · hour256 GB | $0.009 | $0.010 |
| per PFLOP FP16 · hour1,307 TFLOPS | $1.721 | $2.025 |
| per TB/s of bandwidth · hour6.00 TB/s | $0.375 | $0.441 |
| Model | Needs | Smallest config | $ / month |
|---|---|---|---|
| all MiniLM L6 v2Sentence Transformers | 1 GBFP8 | 1× MI325Xat OpenRelay | $1,643 |
| paraphrase multilingual MiniLM L12 v2Sentence Transformers | 1 GBFP8 | 1× MI325Xat OpenRelay | $1,643 |
| bge-m3BAAI | 1 GBFP8 | 1× MI325Xat OpenRelay | $1,643 |
| Qwen3 0.6BQwen | 1 GBFP8 | 1× MI325Xat OpenRelay | $1,643 |
| all-mpnet-base-v2Sentence Transformers | 1 GBFP8 | 1× MI325Xat OpenRelay | $1,643 |
| bge-reranker-v2-m3BAAI | 1 GBFP8 | 1× MI325Xat OpenRelay | $1,643 |
Monthly = 730 hours at the cheapest current on-demand price for that GPU count · models sized at FP8 when available, weights plus runtime overheadSize any model →
MI325X · comparisons and collectionsCompare the MI325X side by side
Questions about renting the MI325X
How much does the MI325X cost per hour?
Today the MI325X rents from $2.25 per GPU-hour on-demand, with a median of $2.65 across 4 providers, each counted once at its cheapest current offer (7 current offers in all, readings of the last 45 days). Prices depend on the provider, the number of GPUs in the machine, the region and whether you rent on-demand, spot or reserved; true cost adds the mandatory add-ons each provider publishes.
Which cloud providers offer the MI325X?
4 providers have a current MI325X offer in stock. Cheapest first: Openrelay ($2.25), Tensorwave ($2.25), Cyfuture Ai ($3.04), Digitalocean ($3.80). Older readings from other providers stay listed on this page, in grey, and are never ranked.
How much memory does the MI325X have?
256 GB of HBM3 at 6.00 TB/s. At FP8 that fits all MiniLM L6 v2 (1 GB), paraphrase multilingual MiniLM L12 v2 (1 GB), bge-m3 (1 GB), Qwen3 0.6B (1 GB), all-mpnet-base-v2 (1 GB) on a single card; larger models need two to eight GPUs or a different card.
Where is the MI325X cheapest right now?
The lowest true cost among eligible offers is $2.25 per GPU-hour at TensorWave for a 8-GPU machine, read within the ranking window. Offers marked unavailable, older than 45 days, superseded by a newer reading, resold or outside the outlier band are listed but not ranked.
Is the MI325X cheaper on spot or reserved?
No fresh spot offer today; reserved terms start at $1.49. Spot suits jobs that checkpoint; reserved suits machines that run all month.